Momento reduces costs and headaches for Web3 startup Goldsky
Goldsky took Momento Cache to production in under one week, unlocking improvements in cost efficiency, reliability, and performance.
Industry: Blockchain, Crypto
Use Case: Caching Blockchain Data
Goldsky is a blockchain infrastructure startup providing real-time data streaming solutions for Web3 applications. We offer a streamlined way for developers to access high-quality blockchain data for their projects, powering transformation, enhancement, and custom data integrations.
Solving cost and reliability problems on the blockchain
The Goldsky Index service architecture depends on both self-hosted and third-party ETH RPC nodes. These upstream endpoints are both unreliable and expensive. We set out to reduce our reliance on these by caching blockchain data. We knew there would be big cost and reliability benefits if we pulled it off—even more if we could avoid the backend operational lift typically required with caching. When we discovered Momento Cache, it seemed like a perfect fit from the start. Totally serverless, supports millions of requests per second (RPS), and integrates easily with our existing architecture.
The first meeting solidified our decision to go with Momento. Quite simply, we really liked their team! We also loved the support we received throughout the PoC process—including advice on what caching strategy would best fit our use case. This is something we knew we needed but never had the bandwidth to really dig into—one of the many reasons why we wanted the serverless operational model.
From dev to prod in under a week
We were able to take Momento to production in less than a week. Getting everything coded up and working actually only took a couple hours—the rest of the time was spent refining our caching strategy to match our particular application requirements.
Momento Cache is now a trusted component deep in our infrastructure. We provide a service called subgraph hosting, which enables users to upload a subgraph to Goldsky which we then manage for them. As part of this, we need to query Ethereum nodes (and other ETH-compatible chains) for block information. We do these queries A LOT. The calls we make to these nodes are expensive within our infrastructure and we cannot control for their reliability, so Momento sits between those upstream providers and our load balancing layer.
During ramp-up, we experienced a few hurdles with timeouts due to resource exhaustion on our Amazon EC2 instances during massive spikes of network traffic. Momento was highly attentive in helping us root cause the issue and added some instrumentation for client-side debugging that allowed us to capture very detailed metrics about our request patterns. They even helped us formulate a horizontal scaling strategy to greatly reduce timeouts. All of this makes us confident about incorporating Momento into future workloads!
Multiplying the benefits
Momento Cache allowed us to reduce the number of outgoing RPC requests our load balancer makes by more than half, while also improving p50 request latency by 75%.
Since deploying, we’ve also gotten reports from users that the Goldsky service indexes 2–5x faster than our competitors—and all of that’s on top of the cost savings we’re seeing internally.
With such an easy implementation, we’re eager to unlock these benefits elsewhere. Right now, we’re investigating how we can use it to add more capabilities to our API server for enhanced functionality for our enterprise users, and have been exploring even more use cases. We plan to use Momento for all our caching needs going forward.
Ready to reduce costs and improve reliability for your applications—and improve performance for your users? Get started with Momento Cache today.
Benny Sidelinger is a Principal Software Engineer at Goldsky. He’s been developing web and mobile applications for over 20 years across companies like Venmo, Bright Machines, and Oracle—to name just a few.
Jeff Ling is the Cofounder and CTO of Goldsky. He’s passionate about building teams and technologies to power the next generation of software development. Prior to Goldsky, he was the VP of Engineering at TrueAccord, the CTO at Uppercase, and one of the earliest engineers at Heap.